Hi Espressif team,
I'm building a personal desk-pet voice assistant (named "Lumi") on ESP32-S3 with esp-sr, and would love a custom WakeNet9 wake word trained via your TTS pipeline (I see TTS Pipeline V3 supports Chinese as of the 2026-04-23 update).
- Wake word (display): 你好露米
- Pronunciation / pinyin (IMPORTANT):
ni3 hao3 lu1 mi1
- 你好 = normal tones (nǐ hǎo)
- 露米 should both be FIRST tone (flat): "lū mī", so it sounds like the English name "Lumi" — NOT the dictionary tones lù (4th) / mǐ (3rd). Please train on the flat
lu1/mi1 pronunciation.
- Target chip: ESP32-S3
- Model: WakeNet9 (wn9); a
_tts (TTS-trained) model is perfect
- Use: personal / hobby project
If it could be added to esp-sr (or shared as a model I can flash to a model partition), that would be amazing. Happy to provide anything else you need. Thank you!
(Related: #88)
Hi Espressif team,
I'm building a personal desk-pet voice assistant (named "Lumi") on ESP32-S3 with esp-sr, and would love a custom WakeNet9 wake word trained via your TTS pipeline (I see TTS Pipeline V3 supports Chinese as of the 2026-04-23 update).
ni3 hao3 lu1 mi1lu1/mi1pronunciation._tts(TTS-trained) model is perfectIf it could be added to esp-sr (or shared as a model I can flash to a
modelpartition), that would be amazing. Happy to provide anything else you need. Thank you!(Related: #88)